作者: Stéphanie Muller , Pascal Lesage , Andreas Ciroth , Christopher Mutel , Bo P. Weidema
DOI: 10.1007/S11367-014-0759-5
关键词: Data mining 、 Probability density function 、 Monte Carlo method 、 Applied mathematics 、 Data quality 、 Computer science 、 Normal distribution 、 Log-normal distribution 、 Distribution (mathematics) 、 Uncertainty analysis 、 Robustness (computer science)
摘要: Data used in life cycle inventories are uncertain (Ciroth et al. Int J Life Cycle Assess 9(4):216–226, 2004). The ecoinvent LCI database considers uncertainty on exchange values. default approach applied to quantify is a semi-quantitative based the use of pedigree matrix; it two types uncertainties: basic (the epistemic error) and additional due using imperfect data). This as implemented v2 has several weaknesses or limitations, one being that always considered following lognormal distribution. aim this paper show how v3 will apply all distributions allowed by ecoSpold data format. A new methodology was developed other than lognormal. consequent formulas were (1) uncertainties combined for distribution (2) links between normal distributions. These points summarized four principles. In order test robustness proposed approach, resulting parameters probability density functions (PDFs) tested with those obtained through Monte Carlo simulation. comparison validate approach. combine distributions, coefficient variation (CV) relative measure dispersion. Formulas express definition each modeling flow its total given. results illustrated values; they agree Some limitations cited. Providing allow assessment (LCA) practitioner select appropriate model datum uncertainty. variability technique can be exchanges also which play an important role v3.